State of the Art and Challenges in Crowd Sourced Modeling

Save this PDF as:
 WORD  PNG  TXT  JPG

Size: px
Start display at page:

Download "State of the Art and Challenges in Crowd Sourced Modeling"

Transcription

1 Photogrammetric Week '11 Dieter Fritsch (Ed.) Wichmann/VDE Verlag, Belin & Offenbach, 2011 Frahm et al. 259 State of the Art and Challenges in Crowd Sourced Modeling JAN-MICHAEL FRAHM, PIERRE FITE-GEORGEL, ENRIQUE DUNN, Chapel Hill ABSTRACT The recent ubiquitous availability of large crowd sourced photo collections (e.g. Facebook, Picassa, Flickr) has opened a wide field of new applications ranging from virtual tourism, cultural heritage preservation to intelligence and defense applications. The research in this active area was energized by the seminal PhotoTourism paper (Snavely et al., 2006) starting from registering 3000 photos on 24 PCs in two weeks. In only five years the community advanced to dense scene reconstruction from 3 million images on as single PC within 24 hrs. At the same time these crowd sourced photo collections are growing at exponential rates posing significant challenges for their long term use. Even the current stateof-the-art methods in scene reconstruction from photo collections lag behind today s demand to scale to the true size of larger city scale datasets like the 12 million images of New York. Even more challenging is the long term demand for world scale reconstructions required for virtual tourism, disaster response and large image based localization for navigation. In this paper we will discuss the goals that need to be reached in term of scalability, robustness against temporal and apperance variation, within the context of generating high quality dense 3D models from heterogenous input imagery. We will present the current state of the art to achieve the goal and some of the challenges that are left to conquer. 1. GOALS FOR CROWD SOURCED MODELLING We identify three crucial performance goals to be achieved by automated crowd sourced modeling systems (CSMS) in order to fulfill current and future performance demands, while broadening the scope of their applicability: Scalable Performance: A CSMS will ideally scale with not more than linear complexity in the number of input images. Such performance would enable city scale reconstructions of 12 million images (number of images of New York on Flickr) or more, while still creating complete models. Robustness to Temporal and Appearance Variation: Large image collections capture complex variation due to physical modification of the scene or atmospheric perturbations that will modify the appearance of a scene being photographed. A reconstruction system needs to be able to cope and detect these variations to ease the scene interpretation. High quality dense 3D models from heterogeneous input: Images found in Internet photo-collection describe the scene very differently because the devices used and their positioning vary greatly. Dense reconstruction systems that use this type of images need to leverage this formidable variation to obtain the most accurate model possible. In this paper we review the state of the art while keeping in mind these goals. We outline some of the approaches that have been presented to meet these challenges and describe some of the open problems that need to be solved in order to obtain a truly usable reconstruction system using crowd sourced modeling.

2 260 Frahm et al. 2. SCALABILITY Figure 1: Performance review of current large-scale reconstruction algorithms. Benchmarking for the work of Frahm et al. (2010) is highlighted in green. The current state of the art algorithms address in parts the challenges of large scale reconstruction. There are two main classes of city scale reconstruction algorithms the first class uses bag of words methods to identify scene overlap, which is then used to bootstrap large-scale structure from motion registration to obtain the spatial camera registration and the sparse 3D point cloud of the scene, see (Snavely et al., 2006 and Agar-wal et al., 2009). These methods were shown to scale to the com-putation of a few hundred thousand images within 24 hrs of computation by using a cloud computer with 64 CPUs. Hence they currently cannot be considered scalable for typical city scale reconstructions, which consist of multiple million images. The second class of methods uses appearance based image grouping followed by epipolar geometry verification to identify overlapping images. Afterwards an iconic scene representation is used to bootstrap the efficient image registration through structure from motion followed by efficient dense reconstruction. These methods can scale to the processing of a few million images on a single PC within 24 hrs as shown in (Frahm et al., 2010). To achieve the scalability these methods compromise model completeness in parts. Both classes of methods fail to achieve location recall beyond the major tourist cites of a city. Bag of words methods: These methods typically provide a higher degree of model completeness in the reconstruction than the appearance clustering approaches. In order to achieve higher completeness they compromise the computational complexity of the method, the seminal Photo- Tourism (Snavely et al., 2006) used exhaustive search to find image overlap. Agarwal et al. (2009) later improved the performance of the approach by using the intrinsic parallelism of the problem to deploy feature based recognition and query expansion. This improved the computational complexity of the overlap detection. Still this only enabled the processing of a few hundred images on a cloud computer (64 PCs). To maintain computational efficiency these methods typically compromise location recall and do not recover scenes captured by a small number of images. Given the latter and foremost the lack of scalability we opt not to develop an approach in this class but rather a novel approach that takes the appearance clustering methods one step further. Appearance based clustering: The strong advantage of the appearance clustering methods is their scalability, which has been shown in (Frahm et al., 2010) to outperform the previous state of the art by at least three orders of magnitude. This is a result of the close to linear computational complexity of the method introduced. A lot of care was taken to limit the number of image comparison. Each image is only compared to a fixed number of candidates. In this way, the first step is to cluster the data in small manageable chunks. While the obtained spatial camera registration delivers the largest so far obtained models of the sites in the reconstructed city, the method compromises on model completeness not achieving as complete models as the bag of words methods due to their more

3 Frahm et al. 261 restrictive overlap search model. Further similar to the bag of words methods, these methods have a small location recall as they do not recover sites covered by a small number of images. Incremental structure-from-motion is the traditional method to create large-scale 3D reconstruction from overlapping images. It is used by algorithms in both of the above classes of methods to obtain the spatial camera registration. Incremental structure-from-motion methods have two main drawbacks. First it is not scalable, as it requires repetitive error mitigation through increasingly larger bundle adjustment, which optimizes the 3D structure simultaneously with the pose of the cameras, see (Snavely et al., 2008). Secondly, it suffers from drift as error is propagated in the reconstruction. This limits the coverage of reconstructed models (Crandall et al., 2011). In order to tackle the problem of efficiency, Ni et al. (2007), proposed to optimize independent submaps in parallel while optimizing the area of overlap in a global thread. This does not only split the problem in parallel units but also in smaller problems that can be optimized more efficiently. Snavely et al. (2008) propose to limit the number of cameras to optimize by using a skeletal set, which provides the same reconstruction accuracy. The skeletal set is determined by searching for the spanning tree of cameras that approximate the uncertainty obtained from the complete set. The major drawback for the computational complexity with the skeletal set is that it still requires a full pair-wise matching of the image connection graph. The approach of Frahm et al. (2010) deploys high-level appearance information to reduce the inherent redundancy in the data by obtaining an iconic scene representation of the photo collection, which represents the scene in a sparse manner, see (Li et al., 2010). Both of the above classes of methods do not address the computational complexity of the error mitigation that is performed through bundle adjustment, which is cubic in the parameter space, nor its increasing numerical sensitivity for a growing number of cameras. In Steffen et al. (2010), the authors developed a relative bundle adjustment based on trifocal constraints, which overcomes both of these limitations. It has the advantages to on one hand limit the size of the state vector, as the 3D structure does not have to be parameterized. This effectively reduces the computational effort. On the other hand it addresses the numerical sensitivity in contrast to traditional approaches by enabling an equivalent not just an approximate differential formulation, which essentially prevents the accumulation of uncertainties that lead to an increasing condition number of the optimization problem. The novel bundle adjustment has shown to have constant uncertainty towards the rotation parameters and linearly increasing uncertainty for the translation due to the unavoidable scale error propagation and it is highly robust to noisy initialization. Recently Crandall et al. (2011) proposed to use a discrete continuous method to limit the number of bundle adjustments. It requires a complicated discrete optimization to initialize the continuous optimizer. Using this method limits the drift drastically and allows for large-scale reconstruction. Strecha et al. (2010) propose to limit drift by registering a set of smaller reconstruction to a street map of the environment. All of the above methods heavily deploy the EXIF information to estimate the focal length along with the other camera calibration parameters like principal point and aspect ratio, which limits the applicability of such methods in the case of crowd sourced imagery as targeted in this proposal. Gherardi et al. (2010) propose a hierarchical reconstruction that runs parallel reconstructions that are iteratively merged. It starts with projective reconstruction and upgrades to a Euclidean reconstruction when the reconstruction contains enough information. By looking at the progress made we can see than scalable solution for matching have been proposed but completeness, high location recall and structure from motion for uncalibrated camera on large photo-collection are still open problems that should be tackled by researchers.

4 262 Frahm et al. 3. ROBUSTNESS TO TEMPORAL AND APPEARANCE VARIATION The role of illumination is dominant in the image generation process. Atmospheric perturbation, time of the day or artificial light will modify the appearance of the picture even if it was taken from the same location with the same camera. Such variations are always present in crowd source photograph because they are captured across time. These photo-collections offer a great source of information to model the captured scene. Unfortunately most algorithm used for dense reconstruction are not invariant to such variation. Goesele et al. (2007) propose two layers selection process. First each candidate images selects neighbors based on features matching, view-angles, and scale changes, and then at pixel level they propose a iterative disparity estimation using region growing starting Figure 2: Example of appearance variation. from triangulated points obtained from structure from motion as they are expected to be more reliable. This method copes well the variation of camera locations presents in reconstruction of photo-collections unfortunately it does not offer a solution for image that have large illumination differences. In order to tackle this challenge we propose to select the images for multi-view stereo based on a clustering using color and gist features. Images presents in each obtained cluster are usually captured under similar condition. This allows us to create dense model from large collection of images capture in an uncontrolled environment. When such a dense model is available, Haber et al. proposed to estimate the reflectance and the incident illumination perimage thus allowing correct texturing of the estimated 3D model. In order to cope with erroneous model resulting from uncertain depth map, Goesele et al. propose to reduce visual artifact by using non-photorealistic rendering for background information. The background pixel color is estimated using random sampling color points along viewing rays. This offers a pleasing ghosting effect that smoothes out artifacts. Color is not the only variation that is present in photo-collections; structural evaluation is also captured by images acquired across time. This is especially the case for 3D reconstruction from cities obtained using images spanning several decades, as new buildings are erected and older ones are destroyed. Schindler et al. propose to infer image acquisition time based on the presence or absence of certain landmark. They do not only estimate a time frame but evolution of a city skyline by estimating temporal visibility probability of each landmark. The goal of handling appearance variation has not yet been reached. Even if we can estimate dense model from varying illumination, every modeling systems only estimate a single texture, thus only representing an average view of the scene where there should be multiple for example switching between day and night.

5 Frahm et al HIGH QUALITY DENSE 3D MODELS FROM HETEROGENOUS INPUT Dense geometry estimation from controlled binocular and multiview stereo sequences is an active and well studied problem. However, multiview stereo (MVS) for CSMS poses the challenge that the assumptions of uniform viewpoint distribution, homogenous sampling scale and scene illumination are not fulfilled across the entire datum. In an effort to mitigate these issues (while maintaining high quality dense models), state of the art approaches have addressed the aspects of intelligent datum segregation while developing their systems around more robust 3D geometric representations. Figure 3: Dense geometry generation. On the right: SfM estimates and the result of image based clustering. Middle: Individual depthmaps obtained for different images in the cluster. At left: Textured 3D model obtained through heightmap based data fusion. View Selection: Global image datum variability (in terms of effective scale, coverage and illumination) can be mitigated by performing partial model reconstruction on image clusters. These clusters aim at grouping images observing the same scene content while sharing common scale and global image appearance. Although such segregation may be directly obtained from preceding SfM modules, it is important to note that favorable viewpoints for sparse structure estimation may not provide an optimum datum for dense reconstruction. Examples of general viewpoint selection schemes for dense reconstruction include the works of Hornung et al. (2008) as well as Dunn and Frahm (2009). Addressing the problem of view selection in the context of crowd source imagery, Goesele et al. (2007) utilized the cardinality of the set of common features among viewpoints along with a parallax metric to build image clusters. Images are then rescaled to the lowest resolution view in the cluster and decisions on image inclusion into the MVS were performed on the pixel level basis according to photo-consistency thresholding. The work in Kanatani (2010) used SfM estimates to discard redundant images while retaining high resolution geometry. In this instance clustering was performed iteratively by merging sparse 3D points and viewpoints based on visibility relations. While no intra-cluster image segregation was made, the cluster size was bounded as a function of available computational resources. In Frahm et al. (2010) it was demonstrated how direct appearance based clustering utilizing color augmented GIST descriptors in conjunction with geometric verification provided a suitable datum for MVS. In Gallup et al. (2008) the concept of variable-baseline and variable resolution was introduced for stereo. This offers an elegant mechanism for heterogeneous image capture integration in MVS systems and may be readily be modified to operate within an image cluster oriented framework instead of the image sequence addressed in the original work.

6 264 Frahm et al. Robust 3D Representations: Variations in ambiance illumination variations and/or partial scene occlusions, are common hindrances to robust dense geometry reconstruction that may not be completely resolved by selective MVS processing. Accordingly, recent work has studied dense structure representations capable of retaining fine structure detail while imposing implicit or explicit regularization on the obtained model surface. All of the following works implemented some form of input data clustering, but differ in the mechanisms used for surface generation. The work of Goesele et al. (2007) implemented clustering followed by region growing mechanisms for dense geometry estimation. In this work, surface normals were controlled explicitly within the photoconsistency function optimization. Recently, Furukawa et al. (2010) was able to implement region growing in city-scale input data. The work of Ackermann et al. (2010) uses MVS techniques to create a partial reconstruction of the scene, which serves as scene intrinsic reference geometry. The authors then transfer surface normals from reconstructed to unreconstructed regions based on robust photometric matching. The work presented in Frahm et al. (2010) addresses surface regularization through the use of a multiple layer heightmap based depthmap fusion module. In this way, water tight models of arbitrary geometry can be efficiently and robustly computed. Moreover, this type of regularization is especially well suited to urban scenes where vertical facades are abundant. 5. OUTLOOK AND CLOSING REMARKS The application 3D reconstruction techniques for large unstructured image datasets is a very active research area that is being fueled by the increasing of availability of crowd sourced data. Current research efforts entail making careful design decisions in order to address the data association problem inherent in these vast photo collections. While the scale of the problem is formidable, a closer inspection also reveals a significant qualitative difference in the input datum. These differences in data scope and scale require also different approaches to integrating 3D reconstruction systems. In this communication we have focused on scalability, robustness and model quality as driving goals and challenges in this relatively new research area. However, suitable benchmarks for all of these goals need to be adopted within the research community in order to perform quantifiable comparisons. 6. REFERENCES J. Ackermann, M. Ritz, A. Stork, and M. Goesele. Removing the Example from Photometric Stereo by Example. In: Proceedings of the ECCV 2010 Workshop on Reconstruction and Modeling of Large-Scale 3D Virtual Environments. S. Agarwal, N. Snavely, I. Simon, S. Seitz, and R. Szeliski. Building rome in a day. IEEE International Conference on Computer Vision (ICCV), pages 1-8, Jul M. Calonder, V. Lepetit, C. Strecha, and P. Fua. Brief: Binary robust independent elementary features. European Conference on Computer Vision (ECCV), IV: , Jul D. Crandall, A. Owens, N. Snavely, and D. Huttenlocher. Discrete-continuous optimization for large-scale structure from motion. CVPR, E. Dunn, and J.-M. Frahm. Next best view planning for active model improvement. In: Proceedings of British Machine Vision Conference, 2009.

7 Frahm et al. 265 P. Fite-Georgel, T. Johnson, and J.-M. Frahm. City-scale reality modeling from community photo collection. ISMAR Workshop on Augmented Reality Super Models, pages 1-4, Oct J.-M. Frahm, P. Fite-Georgel, D. Gallup, T. Johnson, R. Raguram, C. Wu, Y-H. Jen, E. Dunn, B. Clipp, S. Lazebnik, and M. Pollefeys. Building rome on a cloudless day. European Conference on Computer Vision (ECCV), pages 1-14, Jun Y. Furukawa, B. Curless, S. M. Seitz, and R. Szeliski. Towards Internet-scale Multi-view Stereo. IEEE Computer Society Conference on Computer Vision and Pattern Recognition, June D. Gallup, J.-M. Frahm, P. Mordohai, and M. Pollefeys. Variable Baseline/Resolution Stereo, IEEE Computer Society Conference on Computer Vision and Pattern Recognition R. Gherardi, M. Farenzena, and A. Fusiello. Improving the efficiency of hierarchical structureandmotion. Computer Vision and Pattern Recognition (CVPR), 2010 IEEE Conference on, pages , M. Goesele, N. Snavely, B. Curless, H. Hoppe, and S. M. Seitz (2007): Multi-View Stereo for Community Photo Collections. Proceedings of ICCV. M. Goesele, J. Ackermann, S. Fuhrmann, C. Haubold, R. Klowsky, D. Steedly, and R. Szeliski (2010): Ambient Point Clouds for View Interpolation. Proceedings of ACM SIGGRAPH. T. Haber, C. Fuchs, P. Bekaert, H.-P. Seidel, M. Goesele, and H. P. A. Lensch. Relighting Objects from Image Collections. Proc. CVPR, June J. Hays and A. A. Efros. Im2gps: estimating geographic information from a single image. Computer Vision and Pattern Recognition, CVPR IEEE Conference on, pages 1-8, A. Hornung, B. Zeng, and L. Kobbelt. Image selection for improved multiview stereo. In: Proccedings of CVPR, pages 1-8, 2008 A. Irschara, C. Zach, J.-M. Frahm, and H. Bischof. From structurefrom-motion point clouds to fast location recognition. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 1-8, Mar X. Li, C. Wu, C. Zach, S. Lazebnik, and J.-M. Frahm. Modeling and recognition of landmark image collections using iconic scene graphs. European Conference on Computer Vision (ECCV), pages 1-14, Jul Y. Li, N. Snavely, and D. P. Huttenlocher. Location recognition using prioritized feature matching. European Conference on Computer Vision (ECCV), pages 1-14, Sep K. Ni, D. Steedly, and F. Dellaert. Out-of-core bundle adjustment for large-scale 3d reconstruction. IEEE International Conference on Computer Vision (ICCV), R. Raguram and J.-M. Frahm. RECON: Scale-Adaptive Robust Estimation via Residual Consensus. To appear ICCV, A. Saxena, S. H. Chung, and A. Ng. Learning depth from single monocular images. Advances in Neural Information Processing Systems, 18: 1161, 2006.

8 266 Frahm et al. A. Saxena, M. Sun, and A.Y. Ng. Make3d: Learning 3d scene structure from a single still image. IEEE Transactions on Pattern Analysis and Machine Intelligence, pages , G. Schindler and F. Dellaert (2010): Probabilistic Temporal Inference on Reconstructed 3D Scenes. IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR). G. Schindler, F. Dellaert, and S. B. Kang (2007): Inferring Temporal Order of Images From 3D Structure. IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR). F. Schaffalitzky and A. Zisserman. Multi-view matching for unordered image sets, or How do I organize my holiday snaps?. Computer Vision ECCV 2002, pages , N. Snavely, S. Seitz, and R. Szeliski. Photo tourism: exploring photo collections in 3d. ACM International Conference on Computer Graphics and Interactive Techniques (SIGGRAPH), pages , N. Snavely, S. Seitz, and R. Szeliski. Skeletal graphs for efficient structure from motion. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 1-11, May R. Steffen, J.-M. Frahm, and W. Forstner. Relative bundle adjustment based on trifocal constraints. ECCV workshop on Reconstruction and Modeling of Large-Scale 3D Virtual Environments, C. Strecha, T. Pylvanainen, and P. Fua. Dynamic and scalable large scale image reconstruction. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 1-8, Mar 2010.

Building Rome in a Day

Building Rome in a Day Building Rome in a Day Agarwal, Sameer, Yasutaka Furukawa, Noah Snavely, Ian Simon, Brian Curless, Steven M. Seitz, and Richard Szeliski. Presented by Ruohan Zhang Source: Agarwal et al., Building Rome

More information

Towards Linear-time Incremental Structure from Motion

Towards Linear-time Incremental Structure from Motion Towards Linear-time Incremental Structure from Motion Changchang Wu University of Washington Abstract The time complexity of incremental structure from motion (SfM) is often known as O(n 4 ) with respect

More information

Incremental Surface Extraction from Sparse Structure-from-Motion Point Clouds

Incremental Surface Extraction from Sparse Structure-from-Motion Point Clouds HOPPE, KLOPSCHITZ, DONOSER, BISCHOF: INCREMENTAL SURFACE EXTRACTION 1 Incremental Surface Extraction from Sparse Structure-from-Motion Point Clouds Christof Hoppe 1 hoppe@icg.tugraz.at Manfred Klopschitz

More information

Building Rome on a Cloudless Day

Building Rome on a Cloudless Day Building Rome on a Cloudless Day Jan-Michael Frahm 1, Pierre Fite-Georgel 1, David Gallup 1, Tim Johnson 1, Rahul Raguram 1, Changchang Wu 1, Yi-Hung Jen 1, Enrique Dunn 1, Brian Clipp 1, Svetlana Lazebnik

More information

C4 Computer Vision. 4 Lectures Michaelmas Term Tutorial Sheet Prof A. Zisserman. fundamental matrix, recovering ego-motion, applications.

C4 Computer Vision. 4 Lectures Michaelmas Term Tutorial Sheet Prof A. Zisserman. fundamental matrix, recovering ego-motion, applications. C4 Computer Vision 4 Lectures Michaelmas Term 2004 1 Tutorial Sheet Prof A. Zisserman Overview Lecture 1: Stereo Reconstruction I: epipolar geometry, fundamental matrix. Lecture 2: Stereo Reconstruction

More information

Segmentation of building models from dense 3D point-clouds

Segmentation of building models from dense 3D point-clouds Segmentation of building models from dense 3D point-clouds Joachim Bauer, Konrad Karner, Konrad Schindler, Andreas Klaus, Christopher Zach VRVis Research Center for Virtual Reality and Visualization, Institute

More information

MetropoGIS: A City Modeling System DI Dr. Konrad KARNER, DI Andreas KLAUS, DI Joachim BAUER, DI Christopher ZACH

MetropoGIS: A City Modeling System DI Dr. Konrad KARNER, DI Andreas KLAUS, DI Joachim BAUER, DI Christopher ZACH MetropoGIS: A City Modeling System DI Dr. Konrad KARNER, DI Andreas KLAUS, DI Joachim BAUER, DI Christopher ZACH VRVis Research Center for Virtual Reality and Visualization, Virtual Habitat, Inffeldgasse

More information

A unified framework for content-aware view selection and planning through view importance

A unified framework for content-aware view selection and planning through view importance MAURO et al.: VIEW SELECTION AND PLANNING 1 A unified framework for content-aware view selection and planning through view importance Massimo Mauro 1 m.mauro001@unibs.it Hayko Riemenschneider 2 http://www.vision.ee.ethz.ch/~rhayko/

More information

Calibrating a Camera and Rebuilding a Scene by Detecting a Fixed Size Common Object in an Image

Calibrating a Camera and Rebuilding a Scene by Detecting a Fixed Size Common Object in an Image Calibrating a Camera and Rebuilding a Scene by Detecting a Fixed Size Common Object in an Image Levi Franklin Section 1: Introduction One of the difficulties of trying to determine information about a

More information

Towards Internet-scale Multi-view Stereo

Towards Internet-scale Multi-view Stereo Towards Internet-scale Multi-view Stereo Yasutaka Furukawa 1 Brian Curless 2 Steven M. Seitz 1,2 Richard Szeliski 3 1 Google Inc. 2 University of Washington 3 Microsoft Research Abstract This paper introduces

More information

Image Retrieval for Image-Based Localization Revisited

Image Retrieval for Image-Based Localization Revisited SATTLER et al.: IMAGE RETRIEVAL FOR IMAGE-BASED LOCALIZATION REVISITED 1 Image Retrieval for Image-Based Localization Revisited Torsten Sattler 1 tsattler@cs.rwth-aachen.de Tobias Weyand 2 weyand@vision.rwth-aachen.de

More information

ACCURACY ASSESSMENT OF BUILDING POINT CLOUDS AUTOMATICALLY GENERATED FROM IPHONE IMAGES

ACCURACY ASSESSMENT OF BUILDING POINT CLOUDS AUTOMATICALLY GENERATED FROM IPHONE IMAGES ACCURACY ASSESSMENT OF BUILDING POINT CLOUDS AUTOMATICALLY GENERATED FROM IPHONE IMAGES B. Sirmacek, R. Lindenbergh Delft University of Technology, Department of Geoscience and Remote Sensing, Stevinweg

More information

A Learning Based Method for Super-Resolution of Low Resolution Images

A Learning Based Method for Super-Resolution of Low Resolution Images A Learning Based Method for Super-Resolution of Low Resolution Images Emre Ugur June 1, 2004 emre.ugur@ceng.metu.edu.tr Abstract The main objective of this project is the study of a learning based method

More information

Kinect & 3D. Tamara Berg Advanced Mul7media

Kinect & 3D. Tamara Berg Advanced Mul7media Kinect & 3D Tamara Berg Advanced Mul7media Recovering 3D from Images How can we automa7cally compute 3D geometry from images? What cues in the image provide 3D informa7on? Slide credit: S. Narasimhan Visual

More information

Computer Vision - part II

Computer Vision - part II Computer Vision - part II Review of main parts of Section B of the course School of Computer Science & Statistics Trinity College Dublin Dublin 2 Ireland www.scss.tcd.ie Lecture Name Course Name 1 1 2

More information

Fast field survey with a smartphone

Fast field survey with a smartphone Fast field survey with a smartphone A. Masiero F. Fissore, F. Pirotti, A. Guarnieri, A. Vettore CIRGEO Interdept. Research Center of Geomatics University of Padova Italy cirgeo@unipd.it 1 Mobile Mapping

More information

THE PERFORMANCE EVALUATION OF MULTI-IMAGE 3D RECONSTRUCTION SOFTWARE WITH DIFFERENT SENSORS

THE PERFORMANCE EVALUATION OF MULTI-IMAGE 3D RECONSTRUCTION SOFTWARE WITH DIFFERENT SENSORS International Conference on Sensors & Models in Remote Sensing & Photogrammetry, 23 25 v 2015, Kish Island, Iran THE PERFORMANCE EVALUATION OF MULTI-IMAGE 3D RECONSTRUCTION SOFTWARE WITH DIFFERENT SENSORS

More information

Online Environment Mapping

Online Environment Mapping Online Environment Mapping Jongwoo Lim Honda Research Institute USA, inc. Mountain View, CA, USA jlim@honda-ri.com Jan-Michael Frahm Department of Computer Science University of North Carolina Chapel Hill,

More information

ASSESSMENT OF VISUALIZATION SOFTWARE FOR SUPPORT OF CONSTRUCTION SITE INSPECTION TASKS USING DATA COLLECTED FROM REALITY CAPTURE TECHNOLOGIES

ASSESSMENT OF VISUALIZATION SOFTWARE FOR SUPPORT OF CONSTRUCTION SITE INSPECTION TASKS USING DATA COLLECTED FROM REALITY CAPTURE TECHNOLOGIES ASSESSMENT OF VISUALIZATION SOFTWARE FOR SUPPORT OF CONSTRUCTION SITE INSPECTION TASKS USING DATA COLLECTED FROM REALITY CAPTURE TECHNOLOGIES ABSTRACT Chris Gordon 1, Burcu Akinci 2, Frank Boukamp 3, and

More information

Building Rome in a Day

Building Rome in a Day Building Rome in a Day Sameer Agarwal 1, Noah Snavely 2 Ian Simon 1 Steven. Seitz 1 Richard Szeliski 3 1 University of Washington 2 Cornell University 3 icrosoft Research Abstract We present a system that

More information

Interactive Dense 3D Modeling of Indoor Environments

Interactive Dense 3D Modeling of Indoor Environments Interactive Dense 3D Modeling of Indoor Environments Hao Du 1 Peter Henry 1 Xiaofeng Ren 2 Dieter Fox 1,2 Dan B Goldman 3 Steven M. Seitz 1 {duhao,peter,fox,seitz}@cs.washinton.edu xiaofeng.ren@intel.com

More information

Fast Matching of Binary Features

Fast Matching of Binary Features Fast Matching of Binary Features Marius Muja and David G. Lowe Laboratory for Computational Intelligence University of British Columbia, Vancouver, Canada {mariusm,lowe}@cs.ubc.ca Abstract There has been

More information

Accurate Georegistration of Point Clouds using Geographic Data

Accurate Georegistration of Point Clouds using Geographic Data Accurate Georegistration of Point Clouds using Geographic Data Chun-Po Wang Kyle Wilson Noah Snavely Cornell University {cpwang, wilsonkl, snavely}@cs.cornell.edu Abstract The Internet contains a wealth

More information

Parallel Tracking and Mapping for Small AR Workspaces

Parallel Tracking and Mapping for Small AR Workspaces Parallel Tracking and Mapping for Small AR Workspaces Georg Klein and David Murray Active Vision Lab, Oxford This is a PDF of the slides of the talk given at ISMAR 2007 Aim AR with a hand-held camera Visual

More information

Robust Panoramic Image Stitching

Robust Panoramic Image Stitching Robust Panoramic Image Stitching CS231A Final Report Harrison Chau Department of Aeronautics and Astronautics Stanford University Stanford, CA, USA hwchau@stanford.edu Robert Karol Department of Aeronautics

More information

From Single Image Query to Detailed 3D Reconstruction

From Single Image Query to Detailed 3D Reconstruction From Single Image Query to Detailed 3D Reconstruction Johannes L. Schönberger 1, Filip Radenović 2, Ondrej Chum 2, Jan-Michael Frahm 1 1 Department of Computer Science, The University of North Carolina

More information

3D Model based Object Class Detection in An Arbitrary View

3D Model based Object Class Detection in An Arbitrary View 3D Model based Object Class Detection in An Arbitrary View Pingkun Yan, Saad M. Khan, Mubarak Shah School of Electrical Engineering and Computer Science University of Central Florida http://www.eecs.ucf.edu/

More information

Epipolar Geometry Prof. D. Stricker

Epipolar Geometry Prof. D. Stricker Outline 1. Short introduction: points and lines Epipolar Geometry Prof. D. Stricker 2. Two views geometry: Epipolar geometry Relation point/line in two views The geometry of two cameras Definition of the

More information

Character Animation from 2D Pictures and 3D Motion Data ALEXANDER HORNUNG, ELLEN DEKKERS, and LEIF KOBBELT RWTH-Aachen University

Character Animation from 2D Pictures and 3D Motion Data ALEXANDER HORNUNG, ELLEN DEKKERS, and LEIF KOBBELT RWTH-Aachen University Character Animation from 2D Pictures and 3D Motion Data ALEXANDER HORNUNG, ELLEN DEKKERS, and LEIF KOBBELT RWTH-Aachen University Presented by: Harish CS-525 First presentation Abstract This article presents

More information

3 Image-Based Photo Hulls. 2 Image-Based Visual Hulls. 3.1 Approach. 3.2 Photo-Consistency. Figure 1. View-dependent geometry.

3 Image-Based Photo Hulls. 2 Image-Based Visual Hulls. 3.1 Approach. 3.2 Photo-Consistency. Figure 1. View-dependent geometry. Image-Based Photo Hulls Greg Slabaugh, Ron Schafer Georgia Institute of Technology Center for Signal and Image Processing Atlanta, GA 30332 {slabaugh, rws}@ece.gatech.edu Mat Hans Hewlett-Packard Laboratories

More information

Digital Image Increase

Digital Image Increase Exploiting redundancy for reliable aerial computer vision 1 Digital Image Increase 2 Images Worldwide 3 Terrestrial Image Acquisition 4 Aerial Photogrammetry 5 New Sensor Platforms Towards Fully Automatic

More information

FACTS - A Computer Vision System for 3D Recovery and Semantic Mapping of Human Factors

FACTS - A Computer Vision System for 3D Recovery and Semantic Mapping of Human Factors FACTS - A Computer Vision System for 3D Recovery and Semantic Mapping of Human Factors Lucas Paletta, Katrin Santner, Gerald Fritz, Albert Hofmann, Gerald Lodron, Georg Thallinger, Heinz Mayer 2 Human

More information

Incremental Surface Reconstruction. from Sparse Structure-from-Motion Point Clouds

Incremental Surface Reconstruction. from Sparse Structure-from-Motion Point Clouds from Sparse Structure-from-Motion Point Clouds Christof Hoppe, Manfred Klopschitz*, Michael Donoser, Horst Bischof Graz University of Technology * Imaging and Computer Vision Research Group Video Analytics

More information

Improved Billboard Clouds for Extreme Model Simplification

Improved Billboard Clouds for Extreme Model Simplification Improved Billboard Clouds for Extreme Model Simplification I.-T. Huang, K. L. Novins and B. C. Wünsche Graphics Group, Department of Computer Science, University of Auckland, Private Bag 92019, Auckland,

More information

A Prototype For Eye-Gaze Corrected

A Prototype For Eye-Gaze Corrected A Prototype For Eye-Gaze Corrected Video Chat on Graphics Hardware Maarten Dumont, Steven Maesen, Sammy Rogmans and Philippe Bekaert Introduction Traditional webcam video chat: No eye contact. No extensive

More information

Announcements. Active stereo with structured light. Project structured light patterns onto the object

Announcements. Active stereo with structured light. Project structured light patterns onto the object Announcements Active stereo with structured light Project 3 extension: Wednesday at noon Final project proposal extension: Friday at noon > consult with Steve, Rick, and/or Ian now! Project 2 artifact

More information

On Benchmarking Camera Calibration and Multi-View Stereo for High Resolution Imagery

On Benchmarking Camera Calibration and Multi-View Stereo for High Resolution Imagery On Benchmarking Camera Calibration and Multi-View Stereo for High Resolution Imagery C. Strecha CVLab EPFL Lausanne (CH) W. von Hansen FGAN-FOM Ettlingen (D) L. Van Gool CVLab ETHZ Zürich (CH) P. Fua CVLab

More information

How does the Kinect work? John MacCormick

How does the Kinect work? John MacCormick How does the Kinect work? John MacCormick Xbox demo Laptop demo The Kinect uses structured light and machine learning Inferring body position is a two-stage process: first compute a depth map (using structured

More information

Structured light systems

Structured light systems Structured light systems Tutorial 1: 9:00 to 12:00 Monday May 16 2011 Hiroshi Kawasaki & Ryusuke Sagawa Today Structured light systems Part I (Kawasaki@Kagoshima Univ.) Calibration of Structured light

More information

Wii Remote Calibration Using the Sensor Bar

Wii Remote Calibration Using the Sensor Bar Wii Remote Calibration Using the Sensor Bar Alparslan Yildiz Abdullah Akay Yusuf Sinan Akgul GIT Vision Lab - http://vision.gyte.edu.tr Gebze Institute of Technology Kocaeli, Turkey {yildiz, akay, akgul}@bilmuh.gyte.edu.tr

More information

Learning 3D Object Recognition Models from 2D Images

Learning 3D Object Recognition Models from 2D Images From: AAAI Technical Report FS-93-04. Compilation copyright 1993, AAAI (www.aaai.org). All rights reserved. Learning 3D Object Recognition Models from 2D Images Arthur R. Pope David G. Lowe Department

More information

Non-Metric Image-Based Rendering for Video Stabilization

Non-Metric Image-Based Rendering for Video Stabilization Non-Metric Image-Based Rendering for Video Stabilization Chris Buehler Michael Bosse Leonard McMillan MIT Laboratory for Computer Science Cambridge, MA 02139 Abstract We consider the problem of video stabilization:

More information

Fast Image-Based Localization using Direct 2D-to-3D Matching

Fast Image-Based Localization using Direct 2D-to-3D Matching This is a preprint for the paper accepted for publication in ICCV 211. c 211 IEEE Fast Image-Based Localization using Direct 2D-to-3D Matching Torsten Sattler, Bastian Leibe, Leif Kobbelt RWTH Aachen University

More information

Keyframe-Based Real-Time Camera Tracking

Keyframe-Based Real-Time Camera Tracking Keyframe-Based Real-Time Camera Tracking Zilong Dong 1 Guofeng Zhang 1 Jiaya Jia 2 Hujun Bao 1 1 State Key Lab of CAD&CG, Zhejiang University 2 The Chinese University of Hong Kong {zldong, zhangguofeng,

More information

A Short Introduction to Computer Graphics

A Short Introduction to Computer Graphics A Short Introduction to Computer Graphics Frédo Durand MIT Laboratory for Computer Science 1 Introduction Chapter I: Basics Although computer graphics is a vast field that encompasses almost any graphical

More information

ENGN 2502 3D Photography / Winter 2012 / SYLLABUS http://mesh.brown.edu/3dp/

ENGN 2502 3D Photography / Winter 2012 / SYLLABUS http://mesh.brown.edu/3dp/ ENGN 2502 3D Photography / Winter 2012 / SYLLABUS http://mesh.brown.edu/3dp/ Description of the proposed course Over the last decade digital photography has entered the mainstream with inexpensive, miniaturized

More information

TouchPaper - An Augmented Reality Application with Cloud-Based Image Recognition Service

TouchPaper - An Augmented Reality Application with Cloud-Based Image Recognition Service TouchPaper - An Augmented Reality Application with Cloud-Based Image Recognition Service Feng Tang, Daniel R. Tretter, Qian Lin HP Laboratories HPL-2012-131R1 Keyword(s): image recognition; cloud service;

More information

Feature Tracking and Optical Flow

Feature Tracking and Optical Flow 02/09/12 Feature Tracking and Optical Flow Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem Many slides adapted from Lana Lazebnik, Silvio Saverse, who in turn adapted slides from Steve

More information

Color Segmentation Based Depth Image Filtering

Color Segmentation Based Depth Image Filtering Color Segmentation Based Depth Image Filtering Michael Schmeing and Xiaoyi Jiang Department of Computer Science, University of Münster Einsteinstraße 62, 48149 Münster, Germany, {m.schmeing xjiang}@uni-muenster.de

More information

3D Time-Lapse Reconstruction from Internet Photos

3D Time-Lapse Reconstruction from Internet Photos 3D Time-Lapse Reconstruction from Internet Photos Ricardo Martin-Brualla 1 David Gallup 2 Steven M. Seitz 1,2 University of Washington 1 Google Inc. 2 {rmartin,seitz}@cs.washington.edu dgallup@google.com

More information

CS 534: Computer Vision 3D Model-based recognition

CS 534: Computer Vision 3D Model-based recognition CS 534: Computer Vision 3D Model-based recognition Ahmed Elgammal Dept of Computer Science CS 534 3D Model-based Vision - 1 High Level Vision Object Recognition: What it means? Two main recognition tasks:!

More information

Stereo Vision (Correspondences)

Stereo Vision (Correspondences) Stereo Vision (Correspondences) EECS 598-08 Fall 2014! Foundations of Computer Vision!! Instructor: Jason Corso (jjcorso)! web.eecs.umich.edu/~jjcorso/t/598f14!! Readings: FP 7; SZ 11; TV 7! Date: 10/27/14!!

More information

A PHOTOGRAMMETRIC APPRAOCH FOR AUTOMATIC TRAFFIC ASSESSMENT USING CONVENTIONAL CCTV CAMERA

A PHOTOGRAMMETRIC APPRAOCH FOR AUTOMATIC TRAFFIC ASSESSMENT USING CONVENTIONAL CCTV CAMERA A PHOTOGRAMMETRIC APPRAOCH FOR AUTOMATIC TRAFFIC ASSESSMENT USING CONVENTIONAL CCTV CAMERA N. Zarrinpanjeh a, F. Dadrassjavan b, H. Fattahi c * a Islamic Azad University of Qazvin - nzarrin@qiau.ac.ir

More information

Probabilistic Range Image Integration for DSM and True-Orthophoto Generation

Probabilistic Range Image Integration for DSM and True-Orthophoto Generation Probabilistic Range Image Integration for DSM and True-Orthophoto Generation Markus Rumpler, Andreas Wendel, and Horst Bischof Institute for Computer Graphics and Vision Graz University of Technology,

More information

Projection Center Calibration for a Co-located Projector Camera System

Projection Center Calibration for a Co-located Projector Camera System Projection Center Calibration for a Co-located Camera System Toshiyuki Amano Department of Computer and Communication Science Faculty of Systems Engineering, Wakayama University Sakaedani 930, Wakayama,

More information

Spatio-Temporally Coherent 3D Animation Reconstruction from Multi-view RGB-D Images using Landmark Sampling

Spatio-Temporally Coherent 3D Animation Reconstruction from Multi-view RGB-D Images using Landmark Sampling , March 13-15, 2013, Hong Kong Spatio-Temporally Coherent 3D Animation Reconstruction from Multi-view RGB-D Images using Landmark Sampling Naveed Ahmed Abstract We present a system for spatio-temporally

More information

DYNAMIC RANGE IMPROVEMENT THROUGH MULTIPLE EXPOSURES. Mark A. Robertson, Sean Borman, and Robert L. Stevenson

DYNAMIC RANGE IMPROVEMENT THROUGH MULTIPLE EXPOSURES. Mark A. Robertson, Sean Borman, and Robert L. Stevenson c 1999 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or

More information

Edge tracking for motion segmentation and depth ordering

Edge tracking for motion segmentation and depth ordering Edge tracking for motion segmentation and depth ordering P. Smith, T. Drummond and R. Cipolla Department of Engineering University of Cambridge Cambridge CB2 1PZ,UK {pas1001 twd20 cipolla}@eng.cam.ac.uk

More information

Polygonal Approximation of Closed Curves across Multiple Views

Polygonal Approximation of Closed Curves across Multiple Views Polygonal Approximation of Closed Curves across Multiple Views M. Pawan Kumar Saurabh Goyal C. V. Jawahar P. J. Narayanan Centre for Visual Information Technology International Institute of Information

More information

Semantic Visual Understanding of Indoor Environments: from Structures to Opportunities for Action

Semantic Visual Understanding of Indoor Environments: from Structures to Opportunities for Action Semantic Visual Understanding of Indoor Environments: from Structures to Opportunities for Action Grace Tsai, Collin Johnson, and Benjamin Kuipers Dept. of Electrical Engineering and Computer Science,

More information

Character Image Patterns as Big Data

Character Image Patterns as Big Data 22 International Conference on Frontiers in Handwriting Recognition Character Image Patterns as Big Data Seiichi Uchida, Ryosuke Ishida, Akira Yoshida, Wenjie Cai, Yaokai Feng Kyushu University, Fukuoka,

More information

Flow Separation for Fast and Robust Stereo Odometry

Flow Separation for Fast and Robust Stereo Odometry Flow Separation for Fast and Robust Stereo Odometry Michael Kaess, Kai Ni and Frank Dellaert Abstract Separating sparse flow provides fast and robust stereo visual odometry that deals with nearly degenerate

More information

The Big Data methodology in computer vision systems

The Big Data methodology in computer vision systems The Big Data methodology in computer vision systems Popov S.B. Samara State Aerospace University, Image Processing Systems Institute, Russian Academy of Sciences Abstract. I consider the advantages of

More information

Simultaneous Gamma Correction and Registration in the Frequency Domain

Simultaneous Gamma Correction and Registration in the Frequency Domain Simultaneous Gamma Correction and Registration in the Frequency Domain Alexander Wong a28wong@uwaterloo.ca William Bishop wdbishop@uwaterloo.ca Department of Electrical and Computer Engineering University

More information

Human behavior analysis from videos using optical flow

Human behavior analysis from videos using optical flow L a b o r a t o i r e I n f o r m a t i q u e F o n d a m e n t a l e d e L i l l e Human behavior analysis from videos using optical flow Yassine Benabbas Directeur de thèse : Chabane Djeraba Multitel

More information

Teaching Big Data by Three Levels of Projects. Abstract

Teaching Big Data by Three Levels of Projects. Abstract Online-ISSN 2411-2933, Print-ISSN 2411-3123 July 2015 Teaching Big Data by Three Levels of Projects Jianjun Yang Department of Computer Science and Information Systems University of North Georgia jianjun.yang@ung.edu

More information

Photo Sequencing. Tali Basha Yael Moses Shai Avidan

Photo Sequencing. Tali Basha Yael Moses Shai Avidan Tali Basha Yael Moses Shai Avidan ECCV2012 Tel Aviv The Input N images taken from different locations at different time steps Random Order Our Result But Who Cares? Capturing the highlights of a dynamic

More information

Build Panoramas on Android Phones

Build Panoramas on Android Phones Build Panoramas on Android Phones Tao Chu, Bowen Meng, Zixuan Wang Stanford University, Stanford CA Abstract The purpose of this work is to implement panorama stitching from a sequence of photos taken

More information

The Scientific Data Mining Process

The Scientific Data Mining Process Chapter 4 The Scientific Data Mining Process When I use a word, Humpty Dumpty said, in rather a scornful tone, it means just what I choose it to mean neither more nor less. Lewis Carroll [87, p. 214] In

More information

Optical Tracking Using Projective Invariant Marker Pattern Properties

Optical Tracking Using Projective Invariant Marker Pattern Properties Optical Tracking Using Projective Invariant Marker Pattern Properties Robert van Liere, Jurriaan D. Mulder Department of Information Systems Center for Mathematics and Computer Science Amsterdam, the Netherlands

More information

Lecture 19 Camera Matrices and Calibration

Lecture 19 Camera Matrices and Calibration Lecture 19 Camera Matrices and Calibration Project Suggestions Texture Synthesis for In-Painting Section 10.5.1 in Szeliski Text Project Suggestions Image Stitching (Chapter 9) Face Recognition Chapter

More information

A Sensor for Simultaneously Capturing Texture and Shape by Projecting Structured Infrared Light

A Sensor for Simultaneously Capturing Texture and Shape by Projecting Structured Infrared Light A Sensor for Simultaneously Capturing Texture and Shape by Projecting Structured Infrared Light Kiyotaka Akasaka,Ryusuke Sagawa, Yasushi Yagi Institute of Scientific and Industrial Research, Osaka University,

More information

Removing Moving Objects from Point Cloud Scenes

Removing Moving Objects from Point Cloud Scenes 1 Removing Moving Objects from Point Cloud Scenes Krystof Litomisky klitomis@cs.ucr.edu Abstract. Three-dimensional simultaneous localization and mapping is a topic of significant interest in the research

More information

Tracking in flussi video 3D. Ing. Samuele Salti

Tracking in flussi video 3D. Ing. Samuele Salti Seminari XXIII ciclo Tracking in flussi video 3D Ing. Tutors: Prof. Tullio Salmon Cinotti Prof. Luigi Di Stefano The Tracking problem Detection Object model, Track initiation, Track termination, Tracking

More information

Introduction. C 2009 John Wiley & Sons, Ltd

Introduction. C 2009 John Wiley & Sons, Ltd 1 Introduction The purpose of this text on stereo-based imaging is twofold: it is to give students of computer vision a thorough grounding in the image analysis and projective geometry techniques relevant

More information

Situated Visualization with Augmented Reality. Augmented Reality

Situated Visualization with Augmented Reality. Augmented Reality , Austria 1 Augmented Reality Overlay computer graphics on real world Example application areas Tourist navigation Underground infrastructure Maintenance Games Simplify Qualcomm Vuforia [Wagner ISMAR 2008]

More information

An approach to stereo-point cloud registration using image homographies

An approach to stereo-point cloud registration using image homographies An approach to stereo-point cloud registration using image homographies Stephen D. Fox a and Damian M. Lyons a a Fordham University, Robotics and Computer Vision Laboratory, Bronx, NY ABSTRACT A mobile

More information

POSE-SPACE IMAGE-BASED RENDERING

POSE-SPACE IMAGE-BASED RENDERING POSE-SPACE IMAGE-BASED RENDERING Anna Hilsmann 1,2 Philipp Fechteler 2 Peter Eisert 1,2 1 Humboldt University of Berlin 2 Fraunhofer HHI The work presented in this paper is funded by the German Science

More information

The Visual Internet of Things System Based on Depth Camera

The Visual Internet of Things System Based on Depth Camera The Visual Internet of Things System Based on Depth Camera Xucong Zhang 1, Xiaoyun Wang and Yingmin Jia Abstract The Visual Internet of Things is an important part of information technology. It is proposed

More information

Visual Data Combination for Object Detection and Localization for Autonomous Robot Manipulation Tasks

Visual Data Combination for Object Detection and Localization for Autonomous Robot Manipulation Tasks Visual Data Combination for Object Detection and Localization for Autonomous Robot Manipulation Tasks Luis A. Morgado-Ramirez, Sergio Hernandez-Mendez, Luis F. Marin- Urias, Antonio Marin-Hernandez, Homero

More information

Facade Segmentation in a Multi-View Scenario

Facade Segmentation in a Multi-View Scenario Facade in a Multi-View Scenario Michal Recky, Andreas Wendel, and Franz Leberl {recky, wendel, leberl}@icg.tugraz.at Institute for Computer Graphics and Vision (ICG) Graz University of Technology, Austria

More information

COMP 790-096: 096: Computational Photography

COMP 790-096: 096: Computational Photography COMP 790-096: 096: Computational Photography Basic Info Instructor: Svetlana Lazebnik (lazebnik@cs.unc.edu) Office hours: By appointment, FB 244 Class webpage: http://www.cs.unc.edu/~lazebnik/fall08 Today

More information

Isolines: Energy-efficient Mapping in Sensor Networks

Isolines: Energy-efficient Mapping in Sensor Networks Isolines: Energy-efficient Mapping in Sensor Networks Ignacio Solis and Katia Obraczka {isolis, katia}@cse.ucsc.edu Computer Engineering Department University of California, Santa Cruz April 15, 2005 Abstract

More information

Turning Mobile Phones into 3D Scanners

Turning Mobile Phones into 3D Scanners Turning Mobile Phones into 3D Scanners Kalin Kolev, Petri Tanskanen, Pablo Speciale and Marc Pollefeys Department of Computer Science ETH Zurich, Switzerland Abstract In this paper, we propose an efficient

More information

MEng, BSc Applied Computer Science

MEng, BSc Applied Computer Science School of Computing FACULTY OF ENGINEERING MEng, BSc Applied Computer Science Year 1 COMP1212 Computer Processor Effective programming depends on understanding not only how to give a machine instructions

More information

Epipolar Geometry and Stereo Vision

Epipolar Geometry and Stereo Vision 04/12/11 Epipolar Geometry and Stereo Vision Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem Many slides adapted from Lana Lazebnik, Silvio Saverese, Steve Seitz, many figures from

More information

Are we ready for Autonomous Driving? The KITTI Vision Benchmark Suite

Are we ready for Autonomous Driving? The KITTI Vision Benchmark Suite Are we ready for Autonomous Driving? The KITTI Vision Benchmark Suite Philip Lenz 1 Andreas Geiger 2 Christoph Stiller 1 Raquel Urtasun 3 1 KARLSRUHE INSTITUTE OF TECHNOLOGY 2 MAX-PLANCK-INSTITUTE IS 3

More information

Diminished Reality using Multiple Handheld Cameras

Diminished Reality using Multiple Handheld Cameras Diminished Reality using Multiple Handheld Cameras Akihito Enomoto Hideo Saito Graduate School of Science and Technology, Keio University, okohama, Japan Abstract In this paper, we present a system for

More information

Using geometry and related things

Using geometry and related things Using geometry and related things Region labels + Boundaries and objects Stronger geometric constraints from domain knowledge Reasoning on aspects and poses 3D point clouds Qualitative More quantitative

More information

Open Source Computer Vision and Analytics in Support of the UAS and ASPRS Communities

Open Source Computer Vision and Analytics in Support of the UAS and ASPRS Communities Open Source Computer Vision and Analytics in Support of the UAS and ASPRS Communities Heather James, Manager Innovative Applications Computer Vision Group Heather.james@kitware.com (518) 290-6576 Best

More information

3D Interactive Information Visualization: Guidelines from experience and analysis of applications

3D Interactive Information Visualization: Guidelines from experience and analysis of applications 3D Interactive Information Visualization: Guidelines from experience and analysis of applications Richard Brath Visible Decisions Inc., 200 Front St. W. #2203, Toronto, Canada, rbrath@vdi.com 1. EXPERT

More information

Assessment. Presenter: Yupu Zhang, Guoliang Jin, Tuo Wang Computer Vision 2008 Fall

Assessment. Presenter: Yupu Zhang, Guoliang Jin, Tuo Wang Computer Vision 2008 Fall Automatic Photo Quality Assessment Presenter: Yupu Zhang, Guoliang Jin, Tuo Wang Computer Vision 2008 Fall Estimating i the photorealism of images: Distinguishing i i paintings from photographs h Florin

More information

Texture Unmapping Combining Parametric and Non- Parametric Techniques for Image Reconstruction

Texture Unmapping Combining Parametric and Non- Parametric Techniques for Image Reconstruction Texture Unmapping Combining Parametric and Non- Parametric Techniques for Image Reconstruction S. Mills and T.P. Pridmore School of Computer Science & Information Technology University of Nottingham Jubilee

More information

PHYSIOLOGICALLY-BASED DETECTION OF COMPUTER GENERATED FACES IN VIDEO

PHYSIOLOGICALLY-BASED DETECTION OF COMPUTER GENERATED FACES IN VIDEO PHYSIOLOGICALLY-BASED DETECTION OF COMPUTER GENERATED FACES IN VIDEO V. Conotter, E. Bodnari, G. Boato H. Farid Department of Information Engineering and Computer Science University of Trento, Trento (ITALY)

More information

GPS-aided Recognition-based User Tracking System with Augmented Reality in Extreme Large-scale Areas

GPS-aided Recognition-based User Tracking System with Augmented Reality in Extreme Large-scale Areas GPS-aided Recognition-based User Tracking System with Augmented Reality in Extreme Large-scale Areas Wei Guan Computer Graphics and Immersive Technologies Computer Science, USC wguan@usc.edu Suya You Computer

More information

Cees Snoek. Machine. Humans. Multimedia Archives. Euvision Technologies The Netherlands. University of Amsterdam The Netherlands. Tree.

Cees Snoek. Machine. Humans. Multimedia Archives. Euvision Technologies The Netherlands. University of Amsterdam The Netherlands. Tree. Visual search: what's next? Cees Snoek University of Amsterdam The Netherlands Euvision Technologies The Netherlands Problem statement US flag Tree Aircraft Humans Dog Smoking Building Basketball Table

More information

The calibration problem was discussed in details during lecture 3.

The calibration problem was discussed in details during lecture 3. 1 2 The calibration problem was discussed in details during lecture 3. 3 Once the camera is calibrated (intrinsics are known) and the transformation from the world reference system to the camera reference

More information

A technical overview of the Fuel3D system.

A technical overview of the Fuel3D system. A technical overview of the Fuel3D system. Contents Introduction 3 How does Fuel3D actually work? 4 Photometric imaging for high-resolution surface detail 4 Optical localization to track movement during

More information

3D Pipeline Segmentation (planar)

3D Pipeline Segmentation (planar) 3D Pipeline Segmentation (planar) Registration (using lines) 3D PHOTOGRAPHY EXAMPLE Ten scans were acquired of façade of Thomas Hunter Building (NYC) Automatic registration. Each scan has a different color.

More information

EFFICIENT VEHICLE TRACKING AND CLASSIFICATION FOR AN AUTOMATED TRAFFIC SURVEILLANCE SYSTEM

EFFICIENT VEHICLE TRACKING AND CLASSIFICATION FOR AN AUTOMATED TRAFFIC SURVEILLANCE SYSTEM EFFICIENT VEHICLE TRACKING AND CLASSIFICATION FOR AN AUTOMATED TRAFFIC SURVEILLANCE SYSTEM Amol Ambardekar, Mircea Nicolescu, and George Bebis Department of Computer Science and Engineering University

More information